At 13w, an i7 is likely to be over twice as fast as an 8w exynos. I'd say Intel did well here.
This whole "8W Exynos" vs "13W i7" bit Anand started has caused people to draw some really bad conclusions.
First of all, if the device can't run both GPU and CPU at max for a sustained period of time (ie, throttles, as well it should, because this is not a normal use case) then its TDP isn't what you get adding the two together. Intel lets its CPUs momentarily go above TDP before it kicks down the boost scaling or throttling (really pretty much the same thing) just the same. That doesn't mean its "real TDP" is higher, and therefore it makes no sense to claim that the Exynos 5250 has an 8W TDP.
Second, people are lumping together CPU + GPU power consumption then using it evaluate perf/W of the CPU alone. Exynos 5250 is being scolded for being allowed to use up to 4W for its GPU without any real consideration of what the perf/W of that part is (Anand's power comparisons don't include a single GPU perf analysis!). I'm sure that if you let the Y series HD4000 use its full turbo frequency it'll take up a big chunk of the power budget leaving little for the CPU, which is why it's not usually possible.
Finally, Intel's parts have been binned, possibly heavily so - I don't know how cherry picked the Exynos part Anand tested is but if he posts the voltage tables you'll get an idea. It has a big swing in voltage levels between bins.